Date of Award

9-5-2024

Publication Type

Thesis

Degree Name

M.A.Sc.

Department

Mechanical, Automotive, and Materials Engineering

Keywords

Data-Driven Methods;Deep Learning;Fault Diagnosis;Model-Based Methods;Satellites;Stochastic State-Space Tracking

Supervisor

Afshin Rahimi

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

The mission reliability of satellites is critical for ensuring their successful operation in various applications, such as communication, navigation, scientific research, and environmental monitoring. This thesis addresses two significant challenges in the field of Satellite Reaction Wheel (RW) fault diagnosis: data scarcity and the need for real-time fault detection and identification capabilities. The first study proposes a novel approach utilizing the Wasserstein Generative Adversarial Network (WGAN) architecture and a Long Short-Term Memory (LSTM) network to overcome data scarcity and test our proposed data-generative model, respectively. This dual-one-dimensional WGAN-LSTM model generates diverse datasets from limited available data and performs fault detection and identification on satellite RWs. Experimental results demonstrate the model's effectiveness in diagnosing faults with high accuracy. The second study focuses on online fault diagnosis, proposing a hybrid framework for real-time fault detection and identification of a single RW onboard a satellite that capitalizes on both data-driven and model-based methods' strong suits. This framework includes a Markov jump-adjusted RW model, a Markov Jump-Adjusted Particle Filter (MJAPF), and a One Dimensional (1D) sliding window Residual Network (ResNet). The Markov jump-adjusted RW model addresses the limitations of data-driven methods, the MJAPF estimates non-linear RW hidden states under non-Gaussian noise conditions, and the ResNet ensures online diagnosis performance. Experiments showed that the hybrid framework can achieve accurate and timely results, even reaching accuracy rates as high as 99% in low-noise conditions. The proposed MJAPF algorithm proved to be a capable estimation technique. However, the proposed MJAPF and ResNet frameworks were proved to be incompatible due to the gap in their perceptions of fault dynamics but proved effective on their own merits. Overall, this thesis contributes to the advancement of satellite RW fault diagnosis by developing novel methods to tackle data scarcity and real-time fault diagnosis challenges, thereby enhancing the mission reliability of satellite RWs and a wide range of other subsystems as well.

Available for download on Friday, September 05, 2025

Share

COinS